Impacts Of Forest Fragmentation On Pachycondyla Obscurans
The impact of forest fragmentation on genetic diversity of Pachycondyla obscurans, a predatory forest-specialist ant species, was investigated in six forest locations using Amplified Fragment Length Polymorphism (AFLP) fingerprinting. There was little difference in genetic diversity between three large forest locations (two locations in continuous forest, plus one 120,000 ha fragment) and three small forest fragments (151-320 ha). Populations from two locations ~60 km apart in continuous forest showed little genetic differentiation, but genetic differences between all other locations (35-164 km apart), which are separated by oil-palm plantations, were much greater. There was no significant relationship between geographical distance and genetic distance (pairwise FST values) between locations.
Overall, we conclude that the genetic diversity of P. obscurans has not been affected by habitat fragmentation over the past 50 years, but locations separated by forest were more genetically similar than those separated by agricultural areas. This implies that an inhospitable intervening matrix may prevent ant dispersal, and may result in genetic erosion of insect populations in the longer term.Keywords: Forest fragmentation, ants, Pachycondyla obscurans, genetic diversity, Sabah MalaysiaINTRODUCTIONForest fragmentation is recognized as one of the major threats to biodiversity (Laurance et al. 2000; Brook et al. 2002). In tropical regions, fragmentation of forest arises from many causes including land-use changes due to agriculture, urbanisation, and commercial logging.
As a result of fragmentation, previously continuous tracts of forest are divided into fragments, and these fragments become habitat 'islands' isolated within inhospitable landscape matrices. True oceanic islands that are small and isolated are known to support fewer species compared with larger islands (MacArthur & Wilson 1967), and similar patterns are evident in habitat 'island' fragments (Rosenzweig 1995). Forest fragmentation and isolation of remnants is occurring in most tropical regions, often coinciding with the locations of biodiversity 'hotspots' (Myers 2003), and so contributing to global loss of biodiversity. Ecological impacts of forest fragmentation and the resulting isolation of populations leads to changes in the structure of the forest and causes disruption of biological processes and ecosystem functioning within fragments (Didham et al. 1996), thus increasing the risk of extinction of populations (Vasconselos et al.
2006). Fragmentation can result in changes to abiotic and biotic factors, resulting in reduced population sizes within fragments, lower immigration rates of native species, more pronounced edge effects, increased immigration of alien species, and greater incidence of natural and human-caused forest disturbance (Turner 1996). The detrimental impacts of these effects on biodiversity mean there is an urgent need to study the consequences of forest fragmentation on populations of forest species in order to develop more sustainable forest management and effective conservation plans.Forest fragmentation is commonly observed in many tropical regions, and on Borneo recent forest fragmentation has arisen primarily from the conversion of forest into oil-palm (Elaeis guineensis Jacq.) plantations (McMorrow & Talip 2001; Brühl & Eltz 2010).
Insects are one of the most diverse groups of animals on Earth, and are especially diverse in tropical forests, but information on the effects of forest fragmentation on insect populations in this region are limited (Benedick et al. 2006). Previous studies on the responses of insects to tropical forest fragmentation have focused mainly on the effects of forest remnant size and isolation on community composition, and species richness and abundance (Klein 1989, De Souza & Brown 1994, Brown & Hutchings 1997, Didham 1997, Davies & Margules 1998, Kitching et al. 2000, Brühl et al. 2003, Benedick et al.
2006). For example, forest fragmentation affects the composition of ant assemblages due to increased colonisation of small fragments by generalist ants that can persist in the surrounding landscape matrix (Brühl et al. 2003; Schoereder et al. 2004; Bickel et al. 2006; Peters et al. 2009). Other studies on insects have shown that the species richness and diversity of butterflies is positively related to forest remnant size and negatively related to fragment isolation (Benedick et al. 2006). Although species richness is usually reduced in small fragments, these studies on butterflies have shown that small fragments can be important for conserving beta diversity at the landscape scale if small fragments support unique species not found elsewhere (Benedick et al. 2006).Previous studies have also examined changes in genetic characteristics of ant populations in forest fragments in Sabah, Borneo (Bickel et al. 2006).
Such information is important for understanding how fragmentation may lead to local adaptation arising from reduced gene flow between fragments (Hartl & Clark 1989; Templeton et al. 2001). In addition, changes in genetic diversity following fragmentation may lead to genetic erosion and genetic drift in populations in fragmented habitats, which may eventually increase the risk of local extinction for species at these sites (Bickel et al. 2006; Benedick et al.
2007). These previous genetic studies in Sabah on ants and butterflies (Benedick et al. 2007; Bickel et al. 2006) have shown that nucleotide diversity was reduced in fragmented habitats with increasing isolation (butterflies) and decreasing habitat size (ants). However, Benedick et al. (2007) concluded that overall changes in genetic diversity following forest fragmentation were relatively slight in butterflies, which could have been due to the relatively large sizes of forest fragments studied, or due to the short time since fragmentation had occurred.
Lack of genetic changes might also imply sufficient gene flow and dispersal of butterflies among fragments. However, dispersal among fragments was unlikely because the Satyrinae butterfly species studied by Benedick et al. (2007), Mycalesis orseis, is dependent on closed-canopy forest (Hamer et al. 2003) and has not been recorded outside forest habitats or in plantations (Benedick et al.
2006). In the case of leaf-litter ants, Bickel et al. (2006) recorded a significant reduction in genetic diversity following fragmentation in a forest specialist species Pheidole annexus, indicating that forest fragmentation resulted in genetic erosion, although this effect was not observed in a second study ant species. By contrast, Benedick et al. (2007) suggested that isolation of fragments was more important than fragment size for genetic diversity of butterflies.
These two studies in Sabah indicate that different taxa may respond differently to fragmentation. These findings may be related to differences in sizes and placements of habitat fragments being studied, as well as differences among taxa in their ability to occupy matrix habitats and disperse among remaining fragments. Thus there is a need for further investigations of the genetic consequences of fragmentation on species to better understand whether or not genetic erosion is likely to be an important factor affecting species persistence in fragments.Ant assemblages provide an important element to study the effects of forest fragmentation on genetic diversity since they include many habitat-specialist species that are especially sensitive to habitat fragmentation (Boswell et al.
1998). Ants are an ecologically important group of insects in tropical forest due to their high abundance and the important ecological functions they perform as ecosystem engineers (Hölldobler & Wilson 1990). Many ant species have limited dispersal abilities leading to genetic erosion, genetic drift and inbreeding in small populations in fragmented forest (Bickel et al.
2006). Inbreeding may ultimately lead to a higher susceptibility of extinction for populations (Frankham 1998), and some previous studies imply that genetic diversity of ants is reduced following forest fragmentation (Brühl et al. 2003).There are several molecular genetic techniques available for quantifying genetic diversity within populations, and in this study, genetic diversity was measured using Amplified Fragment Length Polymorphism (AFLP) (Vos et al. 1995).
The ability of AFLP to produce bands (DNA fragments) from numerous loci (>1000) at moderate financial costs has made it ideal for evaluating genetic diversity within populations, and for studying population genetic structure (Lynch 2002; Luikart et al. 2003). Some previous studies have used AFLPs for investigating the effects of habitat fragmentation on plants (Stanton et al.
2009) and insects (Monaghan et al. 2002), and in tropical regions they have been used to study the genetic diversity of Aedes aegypti populations in Mexico (Ravel et al. 2001).The present study examined the effects of forest fragment size and isolation on genetic variability of a forest-specialist species of ground-dwelling ant Pachycondyla obscurans (Walker, 1859) from six populations in Sabah (Malaysian Borneo). P. obscurans is a leaf-litter forest specialist species that occurs relatively commonly at forest sites in Sabah. There are few studies on the ecology of P. obscurans, but it is classified as a small-sized species (workers are ~4 mm in length), with nests usually constructed in soil or rotten wood.
The species-group occurs in the Australasian, Oriental and Ethiopian regions (Yamane 2007), and is synonymous with Brachyponera (Bolton 1995), whose workers are small, solitary epigaeic generalist predators and scavengers in forest litter, suggesting the species is likely to be adversely affected by forest fragmentation. Pachycondyla obscurans shows marked intra-sexual dimorphism between workers and queens, with the workers having completely lost their reproductive organs and queens having a large number of ovarioles (Christian Peeters, Laboratoire Ecologie & Evolution, Université Pierre et Marie Curie; pers. comm., 7 January 2011).This ant species was chosen for genetic study because of its restricted habitat range as a specialist predator, and because this species has not been recorded outside forest habitats in the surrounding oil-palm plantations (e.g. Brühl & Eltz 2010).
Therefore, P. obscurans populations in forest fragments are likely to be isolated from one another with limited gene flow between fragments. In this study, we investigated changes in genetic diversity among populations of P. obscurans following forest fragmentation, and we tested the hypothesis that genetic diversity is lower in smaller fragments and related to geographical distance between sites.
MATERIALS AND METHODS
Twenty workers of Pachycondyla obscurans were sampled from each of six study locations in Sabah between June to September 2006 and October to November 2007. Ants were sampled along 2-km transects at each study location. A total of five sites were sampled along each transect using Winkler leaf-litter extraction techniques. Sites along transects were 500 m apart to ensure that samples were of individuals from different colonies. At each site, five 1m2 quadrats were sampled, that were each >10 m apart to ensure that sampled individuals were likely to have come from different nests (Kaspari 1993; Brühl et al.
2003). A maximum of four workers per site (no more than one worker per quadrat) were selected for analysis, resulting in a total of 20 individuals analysed per 2-km transect in each forest location. Our sampling resulted in a total of 120 ant individuals being analysed.
Study locations were in two areas in continuous lowland rainforest (Danum Valley Conservation Area, grid reference 4°55' N, 117°40' E; Maliau Basin Conservation Area, 4°49' N, 116°54' E) and one large forest fragment (Tabin Wildlife Reserve (WR), area = 111,971 ha; Figure 1). These three locations are termed 'large' forests in subsequent analyses. These forest reserves are located in the Lahad Datu-Tawau district and surrounded by oil-palm plantation, with Tabin WR surrounded by especially large tracts[?]. Danum Valley and Maliau Basin are surrounded by secondary forests that were logged ~50 years ago. Samples were also obtained from three 'small' forest fragments (Keruak Virgin Jungle Reserve (VJR; area = 225 ha), Pin Supu VJR (151 ha) and Sungai Sapi VJR (320 ha).
These smaller fragments were protected Virgin Jungle Reserves which supported relatively undisturbed primary forest habitats, and thus were likely to contain forest of similar quality to the large forest sites. Keruak VJR and Pin Supu VJR are located in Kinabatangan District approximately 42-50 km from Sandakan. Both forest reserves are surrounded by oil-palm plantation and small villages.
The forest reserves have a mixture of swamp forest, limestone forest and dominantly mixed dipterocarp forest where timber was selectively extracted ~50 years ago. Sungai Sapi VJR is located in Sandakan-Telupid highway, also surrounded by oil-palm plantation. The forest is mainly lowland dipterocarp forest and partly heath forest.
Secondary forest is commonly found adjacent to the oil-palm plantations. These forests are suspected to have been isolated for ~40 years (McMurrow & Talip 2001). Distances between forests are provided in Table 2. Straight-line distances between locations were measured following the procedure in Benedick et al. (2007).Figure 1. Location of the study sites (black shading) in Sabah, Malaysian Borneo showing the large forest sites (Maliau Basin, Danum Valley and Tabin) and small forest remnants (Sungai Sapi VJR, Pin Supu VJR and Keruak VJR). Dark grey shading shows forest areas, light grey shading shows agricultural areas (i.e. oil-palm plantations) and white shading shows unclassified land-uses (i.e. settlements, open lands etc.).
Source: World Database of Protected Areas; http://www.wdpa.org; Downloaded in 2009).
DNA tissue preparation and extraction
Each worker ant was placed in individual 1.5 ml centrifuge tubes with absolute ethanol stored at room temperature. Subsequently the gaster was removed using a sterilized razor to reduce the amount of formic acid and other inhibitors, and DNA was extracted from head and thorax tissue. The tissue was transferred to new clean tubes with 100 Î¼L extraction buffer containing 20 Î¼L of proteinase K and RNAse A. The tissue was ground with a pestle and incubated at 55oC for 1 h. Once the DNA was extracted, DNA samples were then eluted in Low TE buffer (40 Î¼L of 10 mM Tris, 0.1 mM EDTA, pH 8.0) and stored at 4oC. Extracted DNA was purified using ethanol precipitation. DNA samples were mixed with 4 Î¼L of 0.4 M sodium acetate and 100 Î¼L of ice-cooled absolute ethanol, placed at -20oC for 1 h and then centrifuged for 15 min at 15871 g. The samples were rinsed with 100 Î¼L 70% ethanol for 5 min, centrifuged again at 15871 g for 15 min and air-dried at room temperature.
Finally, the dried DNA pellet was eluted with 20 Î¼L TE and stored at -20oC prior to AFLP analysis.
AFLP analyses were conducted following protocols described by Whitlock et al. (2008A), with slight modifications to ensure that ant DNA inhibitors were reduced. Diluted DNA was subjected to restriction at 37oC for 3 hours using EcoRI and MseI endonucleases. Restricted DNA was ligated at 16oC overnight before diluting with 50 µL sterile water. Preselective PCRs were run using twice the volume of ligated DNA and PCR reaction mix than stated in Whitlock et al. (2008A); 10 µL total reaction volume. To ensure that the DNA was completely free from inhibitors during PCR reactions, the concentration of MgCl2 was decreased (from 50 mM to 25 mM) and BSA (30 mg ml-1) was added.
Diluted pre-selective PCR product was subjected to selective PCR amplification by obtaining 1 µL of the diluted PCR products combined with PCR master mix. Three EcoRI/MseI primer combinations were used to generate the AFLP profiles (EcoRI-TCT/MseI-CGT; EcoRI-TCT/MseI-CGA; EcoRI-TCT/MseI-CTT). Selective PCR products were diluted by a factor of 1:100 and the samples were prepared for genotyping on the ABI 3730 sequencer.
Sequencing preparation protocol
Five µL of ABI Rox size standard was added to 1 ml HiDi formamide (ABI), and a 9.5 µL aliquot of this mix was added to each well of the sequencing plate with 0.5 µL of the diluted selective PCR products. The plate was denatured at 95oC for 3 min and placed on ice afterwards. The sequencing plate was sealed with a septum and then run on an ABI 3730 sequencer using the default settings for AFLP analysis.
Genotyping and AFLP scoring
GeneMapper software (version 3.7) was used to visualise the AFLP fingerprints; fragment size ranged from 50-500 base pairs (bp). GeneMapper was then used to assign bin positions and produce a matrix of peak heights for all the samples at these loci. The resulting peak height matrix was scored for presence or absence of bands using the 'R' script AFLPscore version 1.3 (Whitlock et al. 2008B).
This method removes the subjectivity associated with manually editing AFLP loci, whilst minimising error rates (Whitlock et al. 2008B). The binary genotype table indicating the presence or absence of a peak at each of the retained loci was then used to investigate genetic diversity and population structure.
Genetic diversity and population genetic structure
The proportion of polymorphic loci at the 95% level and the expected heterozygosity (Hj: which is analogous to Nei's gene diversity He) were determined using AFLP-SURV software (Vekemans et al. 2002). Genetic structure was determined using FST and AMOVA (Excoffier et al. 1992) using the Arlequin 3.11 software (Excoffier et al. 2005).
Isolation-by-distance was examined using a Mantel test with AFLP-SURV version 1.0 software.
A total of 120 ant samples was analysed of which 110 samples produced useful data based on output generated from GeneMapper software. The 10 samples that were discarded showed irregular profiles that could not be corrected. Table 1 shows the number of samples analysed from each of the six sample locations.
Genetic diversity and population genetic differentiation
A total of 168 loci were scored from six populations of P. obscurans based on AFLP analysis. There was little difference in the percentage of polymorphic loci recorded at locations following forest fragmentation (small fragments, mean = 79.2% (133/168 loci); large sites, mean = 75.0% (126/168 loci); Table 1). Of all six study locations, Sungai Sapi VJR (area 320 ha) had the highest percentage of polymorphic loci (89.3%) while Pin Supu VJR (151 ha) had the lowest (73.8%). Thus there was little genetic variation observed overall, and the greatest variation was between two of the small forest sites.
The mean expected heterozygosity was also similar between small forest fragments (Hj = 0.253; range 0.241-0.272) and large forest fragments (Hj = 0.246; range 0.193-0.276).Populations of P. obscurans showed significant genetic differentiation based on 1000 random permutations of AFLP data (FST = 0.1614, df = 5, P < 0.05). There was significant genetic differentiation both among populations (FST; each forest fragment) and among populations within the two groups (FSC; forests in large or small fragments), but not between the groups (FCT; large and small fragments) based on a hierarchical AMOVA (AMOVA; FCT = 0.03, df = 1, P = 0.31; FSC = 0.15, df = 4, P < 0.0001; FST = 0.17, df = 104, P < 0.0001). Just 2.73% of the total variation observed was explained by fragment size.
Pairwise genetic differentiation was highest between a large site (Tabin Wildlife Reserve, area = 111,971 ha) and the smallest fragment (Pin Supu VJR, area = 151 ha; Slatkin's linearised FST = 0.7063, df = 5, P < 0.05; Table 2). The least-differentiated populations were in Danum Valley and Maliau Basin (Slatkin's linearised FST = 0.003, df = 5, P > 0.05), which were within the same large tract of continuous forest (Table 2). Sites separated by oil-palm plantations were more genetically differentiated than the two locations linked by forest, and the smallest forest fragment (Pin Supu VJR, 151 ha, Table 2) was the most genetically differentiated of all the study sites.
Overall, there was no significant relationship between geographical distance and genetic distance (pairwise FST values) between study locations (Mantel test; Z = 0.39, P = 0.08; N = 15), although the trend was approaching significance.
There were no differences in genetic diversity of P. obscurans according to fragment size (large versus small locations), contrary to a previous study of ants which reported reduced genetic diversity in fragments (Bickel et al. 2006). However in this previous study, significantly reduced diversity in small fragments was observed in only one ant species (Pheidole annexus; Bickel et al. 2006), with no differences recorded in another ant species (Odontomachus rixosus), implying that genetic effects of fragmentation vary among ant species.
In addition, a study of butterflies in Sabah also found little effect of forest fragmentation on genetic diversity (Benedick et al. 2007). Benedick et al. (2007) sampled in five of the six sites sampled in this study (all but Pin Supu VJR), and the similar results in the two studies imply that fragmentation had relatively little impact on genetic diversity in either of these two insect taxa.
In both these studies, the smallest fragments surveyed were >100 ha, and the relatively large size of the smallest forest remnant sampled may help explain the lack of any significant changes in genetic diversity in either ants or butterflies. Many other studies of habitat fragmentation sampled much smaller fragments (e.g. 20 ha fragment; Bickel et al.
2006) than considered here, and this may account for the reduced genetic diversity they reported. As pointed out by Benedick et al. (2007), even though many butterfly species fail to disperse through agricultural areas, the relatively short time since fragmentation in Sabah (~50 years) and the relatively large population sizes of butterflies in fragments may also contribute to the lack of any genetic erosion being evident within butterfly populations. We have no information on the population sizes of P. obscurans in our study sites, but the fact that the species was commonly recorded at sites during our study implies that the effective population size may be relatively large.
There may also be a time lag between fragmentation and genetic erosion such that longer periods of time (and more insect generations) are required before effects of fragmentation are manifest.The results from this study are likely to be robust and unlikely to be due to a sampling artefact because similar numbers of ants were sampled from each population. In addition, distances between sites along transects mean that samples were likely to be of individuals from different colonies. Nonetheless, only six forest locations were compared and so the analysis may have lacked power in this respect.
The lack of any decline in genetic diversity in small fragments was supported by the similar proportion of polymorphic loci at all study locations (Table 1), also implying little effect of fragment size on ants. Thus we conclude that the populations within the study locations were sufficiently large and/or fragmentation had occurred sufficiently recently that few differences in genetic diversity were present in relation to fragment size. The dispersal ability of P. obscurans is not known and there is no quantitative information on its ability to persist in disturbed habitats. Nonetheless, it is possible that lack of changes in genetic diversity may also be due to sufficient gene flow between sites if reproductive adults can fly between fragments across the intervening oil-palm habitat. Further studies are required sampling ants within agricultural landscapes to investigate this possibility.
However, the lack of any genetic changes in diversity suggests that the smallest forest fragments sampled in this study currently support genetically viable populations of the ant Pachycondyla obscurans.Some studies have hypothesized that genetic differentiation of invertebrates in fragmented forests occurs after ~200 years (Keller & Lagiarder 2003; Keller et al. 2004), and Benedick et al. (2007) suggested that at least 100 generations (years) may be necessary for genetic erosion effects to be evident in butterflies. Time periods taken to observe genetic erosion in ants are likely to be even greater than in butterflies because queens can live up to 40 years (Hölldobler & Wilson 1990).
Moreover, because ants live in colonies where only queens and males reproduce sexually, the effective population size can be small relative to nest size. Thus the length of time for genetic erosion to occur may be much longer than the time-since-fragmentation in this study and may explain why few differences were observed. Nonetheless, Bickel et al. (2006) showed that nucleotide diversity of the generalist ant (Pheidole annexus) was lower in forest fragments in the same study region (Sabah).
This difference between species may have been due to differences in their ecology, such as nesting preferences, colony size, and body size (Bickel et al. 2006). Differences in sizes of fragments sampled, or in the degree of habitat disturbance within fragments or the nature of the surrounding matrix (oil-palm plantation), may also account for these different results, as well as differences in species dispersal ability and habitat associations, and more studies are required.
Population genetic differentiation
The lack of any isolation-by-distance effect in this study may reflect that ~50 years of isolation may not be sufficient to affect dispersal and gene flow among fragments. Pachycondyla obscurans is a forest specialist and has not been recorded in oil-palm plantations, and so this species is unlikely to be able to disperse across agricultural landscapes and genetic differences among fragments may emerge over time assuming that winged adults cannot disperse between fragments. All locations separated by oil-palm plantations were more genetically differentiated than were populations from two locations in continuous forest, supporting the notion that oil-palm habitats prevent barriers to dispersal.
These locations in continuous forest were not in particularly close proximity (~60 km apart) but are linked by forest habitat which may be important for maintaining gene flow and reducing threats from genetic erosion.
To our knowledge, this is the first report of the use of AFLP techniques to study genetic diversity of ants in fragmented tropical forests. Similar studies by Bickel et al. (2006) used Randomised Amplified Polymorphic DNA (RAPD). RAPDs are considered to be less powerful than AFLPs for studying intra-and inter-specific differentiation (Gyzl et al. 2004), implying that our methods would have been more likely to detect an effect of fragmentation if one had been present.Our study showed that genetic diversity of a ground-dwelling ant was not greatly affected by forest fragment size, and ~40 years since fragmentation may not be a sufficiently long time to detect any isolation-by-distance effects. Our findings suggest that forest fragments larger than 150 ha may be of especially high conservation value and may support viable populations of ants.
Two study locations that were in the same block of continuous forest were most genetically similar, even though these sites were 60 km apart. By contrast, ant populations in sites separated by oil-palm plantations were more genetically distinctive, indicating that the nature of the intervening matrix is important. There is an urgent need to evaluate the conservation value of much smaller forest fragments in Sabah, and to examine the degree to which forest species can disperse across agricultural landscapes, such as oil-palm plantations.
Improving landscape connectivity in agricultural landscapes may help promote dispersal of forest species and gene flow among forest fragments.
Article name: Impacts Of Forest Fragmentation On Pachycondyla Obscurans essay, research paper, dissertation